The present invention relates generally to photon cameras and more particularly to multiwire proportional counters and delay line readouts for such counters. The present invention is of particular utility in the field of clinical nuclear medicine imaging.
In clinical nuclear medicine imaging, small quantities of nuclear material are injected into patients, and the location and distribution of the material are determined via the use of external detectors which respond to the radiation emitted from the patient. The systems in current use are based on sodium iodide (NaI) crystal technology whereby the emitted photons are first converted to light within the crystal, and photomultiplier tubes are then used to amplify and locate those faint light emissions.
One such system in current use, developed by Hal Anger, employs a large bank of photomultiplier tubes. In one embodiment, 37 three-inch tubes are used to view a single large circular NaI crystal having typical dimensions of one-half inch in thickness and ten inches in diameter. Each interaction in the crystal is viewed by a large number of the photomultiplier tubes within the bank. Event position is obtained through examination of the relative pulse heights from the various tubes. In Anger's system, position coordinates are obtained by the addition of tube signals in accordance with a weight factor dependent upon the position of the particular tube within the array, thus providing X and Y coordinate signals whose amplitudes are proportional to the X and Y Cartesian coordinates of an event within the crystal.
The Anger system has several significant disadvantages and shortcomings. The lengthy light emission from the NaI crystal limits event acquisition to rates of no greater than about 2.times.10.sup.5 counts per second ("cps") before severe "pile-up" occurs. In pile-up, light from two distinct events enters the photomultiplier tubes at or about the same time, resulting in incorrect determination of position. An additional difficulty is that the discrete nature and the large size of the light detection elements, i.e., the phototubes, produces severe image distortions. These so-called "pin cushion" nonuniformaties must be corrected by complex non-linear transformations. Although this correction can be performed with some effectiveness, significant operational and maintenance problems result because of the need to constantly update the transformation, perhaps on a daily basis, due to such factors as shifts in the photomultiplier tube gain, light coupling efficiency, and shifts in the earth's relative magnetic field.
Additionally, the Anger systems are inherently high energy photon detectors, preferring photons of energies exceeding 80 kev for optimal signal output. These high energies necessitate heavy shielding around the sides and back of the devices as well as heavy collimators (devices used to project images onto the crystal). Consequently, it is not unusual for the heads of these bulky devices to weigh 500 pounds and to occupy three to five cubic feet of space.
Besides the Anger system, a second-type of NaI camera in common use is the multiple crystal camera. Cameras of this design determine position via a matrix of small isolated NaI crystals, each crystal having a facial cross section of about one square centimeter. Each crystal is connected by a split light pipe to two phototubes in such a way that, for example, a 14 by 21 array of crystals can be read out with a 35 tube array. This type of camera is capable of a much higher detection rate, approximately 4.times.10.sup.5 cps vis-a-vis 2.times.10.sup.5 cps, than is the Anger system. However, it is even bulkier and heavier than the Anger type, is expensive as a result of its complexity, has much poorer spatial resolution (1 centimeter) as a result of the size of the crystal matrix elements, and also has very poor ability to measure the energy of the interacting photons, an important requisite in nuclear medicine applications.
The several disadvantages of the sodium iodide crystal cameras leads one to explore other types of devices, especially for use in nuclear medicine, including multiwire proportional cameras. The early detectors in this field were a natural extension of the single wire proportional detector which has been employed for many years for detection and energy measurements of ionizing radiation of many forms. In this detector, a single fine detection wire is mounted within a cylindrical container and is operated at a high positive potential relative thereto. A noble gas (such as argon, krypton, or xenon) is maintained within the intervening space with an admixture of a quench agent, such as methane or carbon dioxide. Ionizing radiation is detected through its interaction in the gas and the consequent release of a small number of free electrons which are amplified through avalanche in the near vicinity of the fine positively charged anode electrode after drift of the free ionization to the region of the anode wire.
Others extended this single wire system to an area detector by providing a planar wire grid anode structure surrounded by an equidistant pair of planar electrodes in place of the cylindrical shell to form cathode electrodes. It was shown that the grid wires in the anode could be placed as close as 1 millimeter apart while still maintaining similar proportional avalanche characteristics to those of single wire detectors, apparently because the cylindrical symmetry holds to a good approximation in areas near each wire where the avalanche phenomenom takes place. Additionally, it was demonstrated that the paired cathode electrodes could be configured in a given grid structure oriented at any angle relative to the anode wire grid and that the cathodes could be employed to determine the position of avalanche on the grid by detecting the induced pulses of equal and opposite magnitude imparted to the cathodes.
While the single wire proportional detector had served as an accurate energy quantity measurement device by virtue of the proportionality between the photon energy input and its signal output, the introduction of the multiwire proportional detector offered great potential for simultaneous position and energy quantity measurement of photons. However, the multiwire proportional counter ("MWPC") also has some disadvantages. One of its disadvantages to date has been its speed. The basic proportional avalanche process is relatively rapid compared with that in other devices such as sodium iodide or spark detectors. Events from a single wire can be recorded at rates exceeding one million per second, and from a large area multiwire detector at rates of up to 100 times this rate, or 10.sup.8 events per second. In contrast, the basic NaI pulse is approximately an order of magnitude slower, limiting rates by a similar factor. Spark devices are slower yet, placing them several orders of magnitude lower in rate compared to the MWPC. Because the MWPC also is capable of equal or superior spatial resolution performance, and is considerably simpler for large area detectors than NaI, the MWPC almost has become the sole means of position measurement in the high energy physics field.
However, heretofore no one has been able to take full advantage of this unique feature of the MWPC (i.e., its ability to perform at very high rates) in many areas of photon imaging. For example, in cardiac imaging a rapidly moving object must be "frozen" with a time resolution of less than one tenth of a second. This means that on the order of 10.sup.5 photons (the number required to form an acceptable image) must be detected during this minimum image formation time. This requires rates of approximately 10.sup.6 photons per second. This rate is theoretically possible with the MWPC, but systems developed heretofore are incapable of producing rates above about 10.sup.5 photons per second.
The devices of U.S. Pat. Nos. 3,772,521 and 3,786,270 to Perez-Mendez and to Borkowski and Kopp are representative of MWPC cameras used in medical (nuclear medicine and radiography) and general materials radiographic imaging. Both describe xenon-filled multiwire detectors substantially similar except for the techniques for electronic digitization of event positions (i.e., their so-called "readout" systems). Each employs three equal-spaced wire planes as detection electrodes, the central plane consisting of fine wires (approximately 20 micrometers diameter) with approximately 2 millimeters spacing, and outer planes consisting of coarser wires (such as 100 micrometers diameter), preferably with orthogonal relative orientation. These wire planes are operated in a gas-tight box which, in the case of the Borkowski and Kopp camera, can be pressurized above atmospheric pressure. The central collection electrode (i.e., the anode) is operated at high positive potential while the surrounding electrodes (i.e., the cathodes) are operated at ground potential. The primary distinction between the two systems is their respective means for decoding the signals in the cathode wires to obtain the position of each photon interaction.
In the Borkowski and Kopp system the cathode wires are connected in a serpentine fashion by resistive elements in such a way that two signals emanating from either end of the serpentine are provided, with each signal's rise time being related to the position of the struck wire within the grid. In particular, the positive ions produced in the avalanche in the vicinity of the anode move toward the cathode planes and induce a displacement current in each cathode which divides into two equal parts and flows through portions of the respective resistor-capacitor (RC) lines and the respective load impedances to ground. The pulse thereby produced across the respective load impedance has a wave shape dependent upon the position of the event within the cathode grids. The pulses are then processed (as taught in U.S. Pat. No. 3,603,797) via timing channel networks comprised of a bipolar pulse shaper and a crossover detector. The output amplitude for each time-to-amplitude converter for each cathode in this system is proportional to the position of an ionizing event on either the X or Y axis as sensed by the particular collector grid.
In the Perez-Mendez system the cathode wires are terminated at one end with a large resistor attached to each wire, and conductor strips are provided at the other end of each wire parallel to the wire and nearly equal in width to the inner wire spacing. Readout is accomplished by placing an electromagnetic delay line in close proximity to those strips, but not in electrical contact with them, so that the wire signals are capacitively coupled into the delay line at a point along its length proportional to the distance of the struck wire at the edge of the detector. Event positions are then encoded by measuring the time delay between the occurrence of a prompting signal obtained from the anode grid and the arrival of signals at each end of the delay line.
As mentioned above, to obtain beneficial use of the MWPC in certain applications, it is necessary to detect up to 10.sup.6 events per second. In order for events to be digitized from a proportional detector operating at that rate, the readout process must be completed within a very short interval, or "readout window" of less than 0.5 microsecond. Also, since photons enter the detector at random times, successive photons may enter within the readout window, thereby producing a confused readout indicative of the actual photon position of neither event. If these pile-up events are not recognized and effectively rejected, they will produce an undesireable background of incorrect photon positions.
Neither the Perez-Mendez nor the Borkowski and Kopp systems meet either of these requirements for beneficial high rate performance. In the capacitively coupled delay lines of the Perez-Mendez system, the optimum delay times of those delay lines are greater than 50 nanoseconds per centimeter. Thus, in a MWPC having a lateral dimension of 25 centimeters, over 1 microsecond is required to clear the signal from a given event from the delay line. The pulse itself from the anode avalanche is almost an order of magnitude shorter than this (approximately 100 nanoseconds). Thus, the inherent readout window imposed by the delay line clearly limits the ultimate rate performance by a similar amount. While this defect might be remedied by reduction of the delay line delay to less than or approximately 10 nanoseconds per centimeter, this is not practical with the capacitive coupling technique inasmuch as it requires high impedence, high inductance lines in order to achieve adequate capacitive coupling efficiency.
The Borkowski and Kopp technique is similarly limited in rate performance. With this technique, the readout window is determined by the RC time constant chosen for the resistive and capacitive readout elements. In order to use variation of rise time as a measure of position, it is clearly necessary that this rise time be large in comparison with the inherent avalanche rise time. Thus, approaching speeds intrinsically allowed by the avalanche process is clearly difficult. In fact, the system described in U.S. Pat. No. 3,786,270 has a rate capability of only about 2.times.10.sup.4 counts per second, far short of the required 10.sup.6 counts per second.
Also, neither the Perez-Mendez nor the Borkowski and Kopp system even purports to address the problem of pile-up. With the Borkowski and Kopp system, it would be particularly difficult to reject pile-up since two events could combine to give an avalanche rise time which would be indistinguishable from an unconfused event. Further, while both systems are capable of partial rejection of pile-up by rejection of anode pulse levels, this method has proven to be inadequate for complete removal of such events.
In addition, other existing systems also have deficiencies which detract from or prevent their usefulness in medical x-ray and gamma ray imaging. For example, a three plane detector has been developed using a variation of the delay line approach, the so-called "integral delay line". However, this system is not useful for most medical or radiography work because its readout is limited to only one coordinate of the Cartesian X, Y pair.